Infrastructure costs only (AWS). OpenAI API costs are billed separately by OpenAI.
- Region: us-east-1 (N. Virginia)
- Team size: ~20 developers
- Average request payload: ~50 KB (request + response log)
- Runtime: Go on
provided.al2023 - Memory: 512 MB
- Average duration: ~20s (depends on OpenAI response time)
- Provisioned concurrency: 2 instances
| Scenario | Requests/month | Lambda Cost |
|---|---|---|
| Low | 20,000 | < $1 |
| Moderate | 60,000 | $1-2 |
| High | 150,000 | $2-4 |
Provisioned concurrency adds ~$5-7/month for 2 instances.
No additional charge (included in Lambda pricing).
- Billing: Pay-per-request (on-demand)
- Operations: 1 GetItem per proxy request
| Scenario | Reads/month | DynamoDB Cost |
|---|---|---|
| Low | 20,000 | < $0.50 |
| Moderate | 60,000 | ~$1 |
| High | 150,000 | ~$2 |
- 1 secret (OpenAI API key)
- Cached in Lambda memory after first call
Cost: ~$0.50/month (fixed).
- Storage: ~50 KB per log entry
- Lifecycle: 90-day expiration
| Scenario | Objects/month | Storage | S3 Cost |
|---|---|---|---|
| Low | 20,000 | ~1 GB | $1-2 |
| Moderate | 60,000 | ~3 GB | $3-5 |
| High | 150,000 | ~7.5 GB | $8-12 |
Costs include PUT requests ($0.005 per 1,000) and storage ($0.023/GB/month).
Lambda execution logs only (not the proxy logs, which go to S3).
| Scenario | Ingestion | CloudWatch Cost |
|---|---|---|
| Low | Minimal | $0 |
| Moderate | ~100 MB | $0-1 |
| High | ~250 MB | $1-2 |
| Scenario | Requests/month | Total (infra) |
|---|---|---|
| Low | ~20,000 (~1k/dev) | ~$2-4 |
| Moderate | ~60,000 (~3k/dev) | ~$6-10 |
| High | ~150,000 (~7.5k/dev) | ~$14-22 |
With provisioned concurrency (recommended): add ~$5-7/month.
- Prices based on us-east-1 as of early 2026; check AWS pricing for your region
- Data transfer out (Lambda -> developer) is minimal and usually within free tier
- The dominant cost for most teams will be the OpenAI API usage, not the proxy infrastructure
- Use
scripts/usage-report.pyto estimate OpenAI costs per developer from the S3 logs